Aviz Logo
AI-Ready Networking Stack
Products which are Open for multiple vendors and AI ready. Driving TCO savings with long term ROI.
Make Networks for AI. Introduce AI in your Networks.
End-to-end solutions — any NOS, any switch, any ASIC, any LLM, any application — backed by partner best practices, proven tech, and SLAs.
Partner Solutions
Celestica
Cisco
Edgecore
Keysight
NVIDIA
Wistron
Explore why Aviz is the best partner to modernize your network with.
Explore Case Studies, TCO and ROI Calculators, Certifications, Community and News room
Aviz Training and Certification
Learn and Certify in SONiC and AI
Why Partner with Us?
  • Open Source Innovation: Lead with SONiC, LLMs, and advanced visibility
  • Network Copilot™ for everyday networking
  • Ready-to-Deploy: Instant IaC playbooks for SONiC
  • Seamless Support: Pre and post-deployment covered
  • Mutual Growth: Scale with shared success
Latest Blog
Latest Blog
Login To Partner Portal
Tailored for Your Role
Explore solutions and tools built for operators, architects, CXOs, and ecosystem partners.
24/7 World-Class SONiC Support & Proven Services
Our dedicated team delivers round-the-clock, world-class SONiC support with unmatched quality, scalability, and efficiency, keeping your network optimized, secure, and always running at its best.
Support
Hamburger
Aviz Logo
Ellipse 1
Hero Background

NVIDIA × Aviz — AI Factory Multi Tenancy with Spectrum-X, Open Networking with SONiC, Service Node Acceleration, and AI-Powered Network Copilot™

From multi-tenant Spectrum-X fabrics to SONiC on NVIDIA Spectrum switches, from DPU-accelerated Service Nodes to Network Copilot™ on NVIDIA GPUs — Aviz and NVIDIA deliver the best software on the best AI networking hardware.

Skip the browsing. Ask AI.
Why This Partnership Matters

NVIDIA builds the most powerful AI networking platforms in the world. Aviz delivers the software stack to make them even better — extracting maximum performance, efficiency, and flexibility. Together, we give customers:

  • Spectrum-X with Multi-Tenant Orchestration and Telemetry — Enable secure multi-tenancy, GPU resource pooling, and AI workload isolation, all with real-time fabric visibility.
  • SONiC on NVIDIA Spectrum Switches — Enterprise-ready community SONiC builds with orchestration, NetOps APIs, 24×7 support, and a back-end SLA with NVIDIA.
  • Open Packet Broker on Spectrum Switches — Harness fully programmable NVIDIA Spectrum ASICs for high-performance, software-first packet processing, filtering, slicing, and tunnel termination.
  • Service Nodes on ConnectX and BlueField DPUs — Offload deep packet inspection and metadata extraction from x86 to DPUs for better performance and power efficiency in high-scale fabrics.
  • Network Copilot™ on NVIDIA GPUs — Deploy private, on-prem AI assistants and networking agents using VLLM or NIMS microservices to deliver intelligent automation and insights..

End of the day, we pair best-in-class hardware with best-in-class software to deliver flexible, scalable, and efficient AI networking solutions.

What You Get
With a single Aviz stack on NVIDIA platforms, customers get five integrated solutions:
Spectrum-X Multi-Tenancy & Telemetry
Multi-tenant orchestration for AI fabrics.
GPU-aware traffic shaping and resource allocation.
Fabric-wide telemetry for real-time monitoring and future capacity planning.
SONiC on NVIDIA Spectrum Switches
Certified SONiC builds are optimized for Spectrum ASICs
Orchestration, visibility, and NetOps integration for multi-vendor fabrics.
Full enterprise support with NVIDIA-backed SLA.
Open Packet Broker on NVIDIA Spectrum Switches
High-speed, line-rate packet processing, slicing, filtering, and tunnel termination.
Leverages Spectrum’s programmable pipeline for advanced observability use cases.
Fully interoperable with Aviz Service Nodes and analytics tools.
Service Nodes on ConnectX & BlueField DPUs
Offload deep packet inspection, deduplication, and telemetry export from CPUs.
Reduce power usage and improve throughput in high-bandwidth deployments.
Ideal for large AI factories and multi-tenant environments.
Network Copilot™ on NVIDIA GPUs
Private AI for networking tasks and automation
GPU-powered inference for large-scale network analytics and agent-based operations.
Deployable via VLLM or NVIDIA NIM.
Where It Works
Use Case
icon1
Data Center Networks
icon1
Edge Deployments
icon1
GPU & AI Fabrics
What It Delivers
SONiC/NVIDIA fabrics with centralized orchestration and real-time analytics.
SONiC-enabled NVIDIA switches for AI/ML workloads at the edge, Kubernetes-ready.
AI Fabric orchestration with SONiC, congestion management, and RoCEv2 support on NVIDIA platforms.
Hardware Compatibility
Procurement ChannelsPort speedPort ConfigSKU
WWT25GbE18x25GbE + 4x100GbESN2010
WWT1GbE48x1GbE + 4x100GbESN2201
WWT400GbE32x400GbESN5600
WWT1GbE48x1GbE + 4x25GbESN2201-48T4X
Proof & Insights
From the NVIDIA Podcast
Joint discussion on AI networking at scale.
From the Spectrum-X Demo
Multi-tenancy orchestration in action.
From the Spectrum-X Webinar
Deep dive into reference architecture integrations.
From SONiC PlugFest
NVIDIA Spectrum SONiC builds validated for enterprise use.
From the Deep Network Observability Interop
Packet Broker and Service Nodes tested with NVIDIA hardware in multi-vendor setups.
From NVIDIA GTC
Live demonstrations of AI networking innovation.
How to Get Started
Start at ONE Center
OCP-recognized lab for testing, validating, and operationalizing NVIDIA-based AI networking solutions.
1
Readiness Assessment — Evaluate your NVIDIA environment for Spectrum-X, SONiC, Packet Broker, Service Nodes, and Network Copilot™.
2
Deployment Planning — Map solution architecture to NVIDIA hardware.
3
Implementation — Install ONES with NVIDIA integrations.
4
Validation — Run at-scale testing for performance, scalability, and observability.
5
Ongoing Support — 24/7 SLA-backed coverage for all NVIDIA × Aviz solutions.
Schedule Your Demo